Conditional Teacher-Student Learning for Model Training

2. The system of claim 1, wherein the trained teacher and untrained student models are associated with at least one of: (i) domain adaptation, (ii) speaker adaptation, and (iii) model compression.

3. The system of claim 2, wherein the trained teacher and untrained student models are further associated with at least one of; (i) a neural network model, and (ii) an acoustic model in an automatic speech recognition system.

4. The system of claim 1, wherein the task is associated with automatic speech recognition and the training data is associated with audio data containing utterances.

5. The system of claim 4, wherein the task is associated with automatic speech recognition domain adaptation of a neural network-based model.

6. The system of claim 5, wherein the trained teacher model is selected based on a selected language.

8. The system of claim 4, wherein the task is associated with automatic speech recognition speaker adaptation of a neural network-based model.

10. The system of claim 9, wherein parameters of the untrained student model are updated according to a back propagation of the student posteriors.

12. The method of claim 11, wherein the task is associated with automatic speech recognition and the training data is associated with audio data containing utterances.

13. The method of claim 12, wherein the task is associated with automatic speech recognition domain adaptation of a neural network-based model.

14. The method of claim 12, wherein the task is associated with automatic speech recognition speaker adaptation of a neural network-based model.

16. The method of claim 15, wherein parameters of the untrained student model are updated according to a back propagation of the student posteriors.

18. The hardware storage device of claim 17, wherein the task is associated with automatic speech recognition and the training data is associated with audio data containing utterances.

19. The hardware storage device of claim 18, wherein the task is associated with automatic speech recognition domain adaptation of a neural network-based model.

20. The hardware storage device of claim 19, wherein the trained teacher model is selected based on a selected language.

22. The hardware storage device of claim 18, wherein the task is associated with at least one of: (i) automatic speech recognition speaker adaptation of a neural network based model, (ii) device personalization providing limited data from a target speaker, (iii) noise speech recognition using clean/noisy speech pair data, (iv) far field speech recognition using close-talk/far-talk speech pair data, (v) kids speech recognition using adults/kids speech pair data, (vi) narrow-band speech recognition using wide-band/narrow-band speech pair data, and (vii) audio-codec speech recognition using original/codec speech pair data.